Subject: Re: A prototype License Wizard up and running
From: "Alex Rousskov" <rousskov@measurement-factory.com>
Date: Thu, 07 Apr 2005 23:48:31 -0600

On Thu, 2005/04/07 (MDT), <jclift@digitaldistribution.com> wrote:

> Alex Rousskov wrote:
>> IMHO, in some cases, there will be more than one "correct answer". We  
>> have  to embrace that instead of trying to "engineer" some of the  
>> answers away.  This is one of the primary differences between CC and  
>> OSI situations.
>>  For example, the wizard might ask:
>>      ``There is no clear legal interpretation of whether X is true.  
>> Even OSI  board is divided on this issue! You can read these  
>> conflicting points of  view <here> and <there>. Do _you_ think X is  
>> true?
>>      - Yes, X is probably true.
>>     - No, X is probably false.
>>     - Hmm... This seems important, but I cannot decide.
>>     - I do not care, why do you ask?!
>> ''
>>  The wizard will then make appropriate adjustments to license  
>> weights... Do  you see what I am getting at?
>
> Sure, it sounds like you want a  
> personality-analysis-slash-expert-system-slash-teaching-thingy.  That'd  
> be cool to have.
>
> Meanwhile, lets write a simple version first... *then* get complicated  
> with it after we've tested the concept.
>
> If you can think of a straightfoward way to implement what you want,  
> then it might be practical in the short term after all.

I believe I can provide low-level technical details on how to implement  
the above with an effort exceeding only an effort required to write a  
script with hard-coded questions and hard-coded decision tree (note that  
the apparent complexity is in the _questions_, not the code "asking"  
them). If it takes a week to implement the above, should we care if a much  
simpler version can be implemented in a day? I do not think so because (a)  
a week is not a big deal when we are 5 years late; and (b) that much  
simpler but rigid version is likely to be discarded almost as quickly as  
it can be written.

Said that, if there is rough consensus that my design is wrong or cannot  
be implemented fast enough, I will simply shut up and gladly wait for  
better alternatives!

Thank you,

Alex.